233 Lancaster Ave · Kitchen nook
Three things changed at once and they compound: a bench down the whole west wall, a custom top instead of whatever a catalogue sells, and a walkway that only has to be squeezable. Together they take this from six crammed places to eight, on 57% more tabletop.
The move
Stop trying to fit a table into the corner and fit a shape that follows the bench. A quarter-ellipse with its corner in the crook — 42″ along the north arm, 54″ down the west — runs its full edge against both benches instead of kissing each one at a single point. That one change is worth four extra seats.
Top
42 × 54 in bean
Places
8 / 6 adults
Tabletop
1781 sq in · +57%
Walkway
20.6 in
01 — Why the shape matters
This is the whole argument. Put a 44″ round in the crook of an L-bench and only about 25″ of each bench arm is close enough to the table to eat at — the rest of your beautiful 62″ bench is a place to put laundry. Give the table two straight edges that run along the bench fronts and the same footprint serves 42″ and 54″.
02 — The layout
The bench runs the full west wall to where the counter starts, and the full north wall to the landing. The table’s corner sits in the crook, its two straight edges run along the bench fronts, and its curved edge faces the room where the chairs go.
03 — How hard to push
You said the path only has to be squeezable and there’s a second way round through the peninsula. That buys real table — but it spends fast, so here is where each size lands.
| Top | Walkway | North arm | West arm | Places | Area | |
|---|---|---|---|---|---|---|
| 38″ round | 27.7″ | — | — | 6 crammed | 1134 | today |
| 44″ round, L-bench | 14.8″ | 25″ | 25″ | 5 | 1521 | the wrong shape |
| fan, 44″ radius | 23.9″ | 44″ | 44″ | 6–8 | 1521 | safest path, symmetric |
| bean 42 × 54 | 20.6″ | 42″ | 54″ | 8 | 1781 | recommended |
| bean 44 × 58 | 17.8″ | 44″ | 58″ | 8–9 | 2004 | only if you truly never carry anything through |
The asymmetry is free money: your west wall runs 77″ and the north only 62″, so stretching the west arm adds seats and tabletop while costing almost nothing on the path — the pinch is on the diagonal, and the west arm grows away from it.
04 — Build spec
One continuous L carcass. Backless, because your 20.5″ sill means a backrest would cover glass. Held 3″ off the wall so the baseboard radiators keep working — that gap is the whole trick and it costs you nothing, because the cushion covers it.
L carcass
Build it as one continuous L so the corner has no seam. Keep the storage in the front 12″ with a full-height divider at the back — nothing should sit against a live convector.
The corner seat
Round the corner, don’t chamfer it. A square point in the crook is the worst seat in any banquette — the two people either side arrive knees-first at the same place, and it’s a 90° point at toddler head height. A tangent arc fixes that while giving up less tabletop and staying closer to the corner seat than a flat facet does.
Start at an 8½″ radius. It is the one edge you can enlarge later in three minutes, so commit to it last — on the mock-up and on the real top. Above about 17″ it stops reading as a rounded corner and starts flowing into the main curve, which is a look worth seeing in plywood before you buy it in oak.
Custom top
A CNC shop cuts this from a drawing in an afternoon. Plywood or MDF core with a solid hardwood edge is the sane build; solid stave-and-shape costs more and moves more.
Base
A low cast disc, not legs and not a trestle. Feet rest over a flat disc; they collide with everything else. Put the column at the top’s centre of area so the longest overhang is about 36″ — have the fabricator confirm tip load at that reach before you commit.
A sheet of plywood, a jigsaw, and an afternoon. By dinner you are eating off the real shape and you will know within one meal whether the bean is right — which is the entire point, and cheap at the price.
Step 1 — the run
Under $150 if you own the jigsaw and a sander. The two items people skip and regret are the reverse-tooth blade and the foam sheet — the first decides whether your good face tears out, the second decides whether cutting a 4-foot curve is pleasant or a wrestling match.
| Item | Why | ~$ | |
|---|---|---|---|
| 1 | 3/4″ sanded plywood, 4×8 | birch if they have it — you’re eating off this | 55–75 |
| 1 | Jigsaw blades, reverse-tooth | Bosch T101BR or similar. Cuts on the down stroke so the top face stays clean | 8 |
| 1 | 2″ rigid foam insulation, 4×8 | lay the sheet on it and cut right through — full support, no pinching, no sawhorse dance | 30 |
| 1 | 1/4″ × 1½″ × 8′ lattice or screen bead | the batten you spring through the dots to fair the curve | 8 |
| 1 | Small brad / wire nails | pin the batten at each dot | 4 |
| 1 | Sandpaper — 80, 120, 150 | 80 to fair, 150 to finish. Get a sponge block too | 12 |
| 1 | Zinsser Bulls Eye shellac, qt | dries in 45 min — the only finish that gets two coats on before dinner | 18 |
| 1 | Cheap 2″ brush or a rag | shellac wipes on fine | 4 |
| 1 | 2×4, 8 ft | three hook spacers — you need 18″ of it, and you’ll want more for the benches | 5 |
| 1 | Cedar shim pack | snugs the hooks up under the stone | 4 |
| 1 | Wood screws, 2″ and 1¼″ | small assorted box. 2″ goes through the spacer and bites the ply without popping through | 9 |
| 3 | F-clamps, 4″ throat (or skip) | the fast alternative to hooks — see the two options below | 30 |
| 1 | Non-slip rug pad or shelf liner | between plywood and stone — no scratches, no sliding, no clamps needed | 8 |
| — | Glasses, dust mask, ear plugs | a jigsaw in plywood for 40 minutes is loud and dusty | 10 |
Home Depot will rip a sheet on their panel saw, and 42 × 54 fits a car that a full 4×8 does not. But the way you ask matters more than the cut does.
Say this at the saw
“Two cuts please — 54¼ inches off the long side and 42¼ inches off the short side, and I need to keep the factory corner.”
Two things are going on there. The quarter-inch is deliberate — panel saws run ±⅛″ on a good day and they don’t promise better, so cutting slightly over means a sloppy cut can never land you undersize. And the factory corner is the whole game: the 42″ and 54″ straight edges are the finished edges of your table, the ones that sit against the benches, and they have to be genuinely straight and square to each other. The factory edges already are. The two cuts they make land on the sides the curve eats anyway, so their accuracy genuinely does not matter.
Step 2 — about an hour
Lay the sheet good-face-up on the foam. Measure along the two factory edges, drop a dot at each offset, pin a batten through the dots and trace. Check square first with a 3-4-5: 36″ along one edge, 48″ along the other, and the diagonal between those marks must read exactly 60″.
Round it, don’t chamfer it — and start at 8½″
A curve beats the flat facet on every count. Off the same marks it pulls the table back only 3½″ instead of 6″, so it stays closer to whoever’s in the crook; it costs 16 square inches instead of 36; and it leaves no corners at all, where a chamfer just swaps one 90° point for two 135° ones. It’s also easier to cut — a jigsaw fairs a gentle arc more happily than it holds a 12″ straight line.
Start small because this is the only edge on the part where being wrong is recoverable. The two straight edges and the main curve can never be added back; the corner enlarges in three minutes any time.
Table A — from the 42″ edge
Measure across the 42″ edge, then down parallel to the 54″ edge.
| across | down |
|---|---|
| 0″ | 54″ |
| 3″ | 53-7/8″ |
| 6″ | 53-1/2″ |
| 9″ | 52-3/4″ |
| 12″ | 51-3/4″ |
| 15″ | 50-1/2″ |
| 18″ | 48-3/4″ |
| 21″ | 46-3/4″ |
| 24″ | 44-3/8″ |
| 27″ | 41-3/8″ |
| 30″ | 37-3/4″ |
| 33″ | 33-3/8″ |
| 36″ | 27-7/8″ |
Table B — from the 54″ edge
Measure down the 54″ edge, then across parallel to the 42″ edge.
| down | across |
|---|---|
| 0″ | 42″ |
| 3″ | 41-7/8″ |
| 6″ | 41-3/4″ |
| 9″ | 41-3/8″ |
| 12″ | 41″ |
| 15″ | 40-3/8″ |
| 18″ | 39-5/8″ |
| 21″ | 38-3/4″ |
| 24″ | 37-5/8″ |
| 27″ | 36-3/8″ |
Corner: mark 8½″ in from each edge. Where the two lines cross is your centre — nail, swing an 8½″ arc.
Step 3 — 40 minutes
Sheet flat on the foam, on the ground. The blade runs into the foam, the panel is supported everywhere, nothing pinches and nothing drops. It is by a distance the easiest way to jigsaw a big curve, and the foam sheet survives to be cut on again.
Why reverse-tooth
A normal jigsaw blade cuts on the up-stroke, so it tears the top face — the one you eat off. A reverse-tooth blade cuts downward and leaves the top clean, which lets you mark and cut with the good face up where you can see your line.
It does want to lift the work slightly. On foam, with your weight on the panel, that’s a non-issue.
If you skip the foam
Two sawhorses and you’ll spend the cut repositioning to keep the offcut from sagging and pinching the blade — and the last few inches will snap off and tear the face. If you go this route, support the waste side too, and slow right down at the end.
Step 4 — an hour, then wait
Fair the curve
Sight down the edge. Plywood shows every wobble as a shadow line, so trust your eye over the pencil. Knock the high spots with 80 on a block — a stiff block, not your fingers, or you’ll sand the dips deeper. Then 120, then 150.
Ease the edge — don’t skip this
Break the sharp arris top and bottom with 120 held at 45°, then round it over. Aim for something you’d happily run a forearm along. A raw plywood edge at toddler head height is genuinely sharp, and you have a 1-year-old.
Shellac, because it’s fast
Wipe or brush a thin coat, wait 45 minutes, scuff with 220, second coat. Touch-dry in half an hour, fine for dinner. Polyurethane would want overnight and you don’t have overnight.
Hit the cut edge harder than the face — bare plywood edge drinks liquid straight into the core.
Belt and braces
Two coats of shellac shrugs off a spilled drink for a few minutes, not a whole dinner with four kids. Put a cloth or a clear vinyl runner down for the meal itself. You’re testing a shape, not a finish.
Step 5 — 20 minutes
Straight on top of the travertine, on a non-slip pad. That alone stops it sliding, and if you want it properly fixed there are two good ways below — but keep the props either way: fixing the top down stops it moving and does nothing about tipping. But the bean is a lot bigger than a 38″ round, and the two tips hang 11″ and 21″ past the edge of the old top — so they need propping. This is the one part of today that’s a safety question rather than a taste question.
The stone only just fits — position matters
The biggest circle this bean can swallow is 38.1″. Your table is 38.0″. So it can be completely covered, but at essentially one position with about a twentieth of an inch to spare — slide it wrong and stone shows.
The rule is easy to hit by eye: the rim just kisses both straight edges, 19″ in from each. It has to, because the stone’s radius is 19″. Get those two right and the curve looks after itself.
The cost: the pedestal ends up about 4″ off the bean’s balance point. Harmless on its own — 32 lb of plywood against 210 lb of table won’t shift it — but the west tip now hangs 21″ past the stone instead of 17″, so that prop earns its keep.
Lift the travertine off and bolt the plywood straight to the pedestal. It solves the covering problem by deleting it, puts the top back on its balance point, needs no clamps or hooks, and is simply how a table is supposed to be built. Go look underneath before anything else.
The catch: the stone was doing real work
That slab is 100–150 lb of ballast, and taking it off roughly halves what it takes to tip the table. Loaded on the west tip it goes from about 81 lb to somewhere near 32–58 lb, depending on what the bare pedestal weighs.
That is squarely in three-year-old-hanging-off-the-corner territory. Do it if it unbolts — it’s the right mount — but the props stop being optional. Alternatively pile roughly 110 lb onto the pedestal foot, which is to say: put the stone back.
Before you commit
Get under there with a torch. A wooden top is usually screwed to a plate. A stone top is very often siliconed to it instead, sometimes with no fasteners at all. Silicone means a wire saw, a mess, and re-gluing later.
If there are bolts into inserts in the stone, note the pattern — you’ll drill the plywood to match, or just run new screws up through the plate’s holes into the ply.
Handling the slab
Two people, and store it flat. Around 120 lb of travertine at 38″ across. On its edge it chips; leaned against a wall it can crack under its own weight or get knocked by a kid. Flat on the floor on a blanket, somewhere out of the traffic, for two days.
This is still the table you want back afterwards — the risk of this option is to the good top, not to the mock-up.
Before anything else: set the plywood on the stone, reach underneath with a pencil and trace the stone’s rim onto the plywood’s underside. Everything here gets easy once you can see where the stone actually is.
Option A — clamps, 5 minutes
Positioned right, the rim comes flush with the plywood edge at three points — 19″ along the 54″ edge, 19″ along the 42″ edge, and once on the curve. A clamp at any of those needs almost no throat.
The catch: the top jaw lands on the tabletop, right where a place setting goes. For a dinner test that’s a real nuisance — three clamp heads among the plates.
Option B — hooks, 20 minutes
Spacer screwed to the plywood underside just outside your traced rim; catch screwed under it, reaching back in; shims driven between catch and stone. Entirely underneath — nothing on the eating surface.
Measure the stone first. A 2×4 laid flat drops 1½″, which suits a stone up to about 1¼″ with shims to spare. Thicker than 1½″ and you want the 2×4 on edge instead.
Say this to the kids once
Don’t lean on, sit on or climb the pointed ends. A propped plywood top is fine to eat off and not fine to treat like furniture — it isn’t fastened down and the tips are cantilevered. Two days, then it comes off.
Step 6 — the actual point
Carry a full laundry basket from the deck door to the kitchen. Twice. This is the 20.6″ pinch, and it’s the thing most likely to make you regret the bean. If it grates, the symmetric 44″ fan gives you 23.9″ back for two seats’ worth of tabletop.
Eat a real dinner on it in the actual positions — three along the 42″ edge, two down the 54″, one at the chamfer. Watch whether the corner seat is genuinely usable or just usable on paper.
Leave it up two days. Mock-ups flatter themselves on day one; the complaints show up on day two. Tape the bench footprints on the floor too — 62″ and 61″, 16″ deep — or you’re only testing half the idea.
After today
Monday — CNC
“One flat part, quarter-ellipse about 42 by 54 with a chamfered corner, 3/4″ ply I supply, outside profile only, I’ll send a DXF — quote cut-only, and what’s the price for a tweaked version of the same file?”
Rates run $60–150/hr, your part is ~14 linear feet of profile. $300–500 is realistic for cut-only on your own sheet. Ask whether they’ll cut customer material — some won’t, hidden staples wreck tooling.
Monday — base
Search “cast iron disc table base, dining height, 30 inch”. The good ones weigh 67–72 lb; the 28 lb cross bases aren’t it. Column is usually 28″, so with a 3/4″ top you land near 29″.
Know the limit: the rule of thumb is under 12″ of overhang past the base edge, and this top reaches 34″ from any single point. Fine for a plywood test top. Not fine for the finished one.
This week — rough benches
2×4 ladder frames with a 3/4″ ply deck: eight uprights at 14¼″, rails at 62″ and 61″, decks 13″ deep. That puts the deck at 15″ and the cushion top at 18″. About $100.
Set them 3″ off the wall, even in the mock — that gap is what tells you whether the radiators still work with something in front of them.
A month — the real thing
White oak at 1½″ is 72 lb; granite at 1¼″ is 218 lb and turns the base into an engineering project. A porcelain slab gets the stone look at 77 lb. Whatever the top, the base is a custom low steel plate cut from the same DXF — it’s the one piece nobody sells off a shelf.
Everything for today, in order, self-contained — you shouldn’t need the other tabs once you start. Plywood sits on top of the stone; unbolting is tomorrow’s problem. Home at 9:55, finished around 3:30, which leaves two and a half hours of slack before dinner.
One reframe before you start
You said cutting, sanding and staining were the hard parts. Cutting is the quick bit — about 35 minutes. The slow one is marking, at roughly 45, and it’s the only step where a mistake is permanent. Everything downstream just follows the line you draw. Go slowly there and fast everywhere else.
| Time | Step | Mins | |
|---|---|---|---|
| 9:55 | Unload | 10 | |
| 10:05 | Foam bed, pick the good face, find the factory corner | 15 | |
| 10:20 | 3-4-5 square check | 5 | |
| 10:25 | Plot the offsets — 24 dots | 25 | slowest, most important |
| 10:50 | Batten and trace the curve | 15 | |
| 11:05 | Swing the corner arc | 5 | |
| 11:10 | Cut it | 35 | |
| 11:45 | Test fit on the stone, trace the rim | 15 | before you sand |
| 12:00 | Fair the edge, 80 grit | 25 | |
| 12:25 | Sand faces, ease edges, round the tips | 25 | |
| 12:50 | Vacuum and wipe down | 10 | |
| 1:00 | Shellac, coat 1 | 15 | |
| 1:15 | while it dries — build the three hooks | 45 | free time |
| 2:00 | Scuff with the sponge, shellac coat 2 | 20 | |
| 2:20 | while it dries — position the table, set the props | 30 | free time |
| 2:50 | Coat 3 if you want it | 15 | optional |
| 3:05 | Mount, fit hooks, tip test | 30 | |
| 3:35 | Done | — | 2½ hrs before dinner |
The two while it dries blocks are the trick. Shellac needs 45 minutes between coats and you have exactly two jobs that don’t touch the tabletop — building hooks and moving furniture. Do them then and the drying costs you nothing.
Step 1 · 10:05 · 15 min
Work on the floor, not on sawhorses. The foam supports the whole panel so nothing sags, pinches the blade or snaps off at the end of the cut.
If you got two foam sheets
Stack them. 3″ total, done. A jigsaw blade swings 2½–3″ below the work at the bottom of its stroke, which is the whole reason for the thickness.
If you only got one
Cut one piece about 44 × 56 as the top layer — that’s what the plywood rests on, so it must be full size and flat. Cut the rest into strips and lay them underneath as sleepers, with gaps. Total 3″.
The blade nicking a sleeper where the curve crosses one is fine. It’s foam. Don’t try to route around your cut line.
Pick the good face
Sanded ply has a better side. Choose it, put a big pencil X on the back, and work with the good face up all day — your reverse-tooth blades cut on the down stroke so the top face stays clean.
Find the factory corner
Two of your edges are Home Depot’s cuts and two are original. The factory corner is where the two uncut edges meet — look for the smoother, slightly rounded, unchipped edges. Mark it with an arrow. Those two become the 42″ and 54″ edges of the table.
Step 2 · 10:20 · 5 min
From the factory corner, mark 36″ along the 42″ edge and 48″ along the 54″ edge. The diagonal between those two marks must read exactly 60″.
If it’s off by more than about 1/8″, you’ve either picked the wrong corner or one of those edges is a store cut. Check again before you plot a single dot — every offset below is measured from these two lines, so a skewed baseline puts a twist through the whole curve.
Step 3 · 10:25 · 25 min
What a “dot” actually is — and no, you don’t need a level
The sheet lies flat on the foam and every number below is measured on that flat surface. Nothing is plumb, nothing is level, gravity plays no part. This is graph paper, not framing.
Each dot takes two measurements, not one. You are not marking one number on one edge and another number on the other edge. Both numbers describe a single point: the first walks you along an edge, then you turn 90° and walk into the sheet by the second. The tick on the edge is scaffolding — the dot itself lands out in the field of the plywood.
Worked example, row two of Table A — 3″ / 53-7/8″: from the factory corner go 3″ along the 42″ edge and tick it. From that tick go 53-7/8″ into the sheet, parallel to the 54″ edge. That is the dot.
Measuring 53-7/8″ square across a sheet is miserable and it is where error creeps in. You never have to: every one of the 24 dots sits close to the two store-cut edges, so work in from those. They are already 42¼″ and 54¼″ from the factory edges — near enough to the lines you would otherwise draw that drawing them is not worth the time.
Name the two edges, then check them once
Edge S is the right-hand cut, edge T the bottom cut. Read the sheet’s width near each end: if both say the same thing, the cut is parallel to the factory edge and you are done checking. Same for the length. Thirty seconds, no lines.
Only if a reading is off by more than about 1/4″ end-to-end is it worth striking your own line — mark the same distance near each end of the factory edge and join the marks.
What the quarter inch actually costs you
Nothing, and it slightly helps. The bean comes out 42¼″ × 54¼″ instead of 42 × 54, and the dots land within 1/9″ of a true ellipse of that size — less than the batten’s own fairing, since a sprung batten physically cannot bend into a wiggle and averages your dots on the way past.
The one number worth knowing: the stone needs 38″ of cover and the bigger bean holds a 38.32″ circle instead of 38.11″. That takes the tightest fit on the whole build from 0.11″ of margin to 0.32″. The lazy version is the better one here.
Now every dot is a small step in from S or T, and no offset in the layout is more than 12-15/16″ — fifteen of the twenty-four are under 5″. That is short enough that squareness stops being delicate: at the very worst dot, being 5° out of square costs you 1/20″, and 5° means the tape is visibly leaning an inch sideways. You are not going to be that far off.
So you don’t need a framing square. In rough order of what is worth reaching for: the plywood offcut — whatever came off the sheet at the store has a factory corner, and it is a bigger and truer square than anything you would have bought; a hardcover book or a cabinet door; or a sheet of paper folded twice, fold onto fold, which gives a genuinely exact 90° for free. Failing all of those, hook the tape on the edge and swing it a little — the smallest reading is the perpendicular one.
Off edge T — the flat run
Tick along edge T, the bottom cut, starting at the factory corner end. Then square in from T.
| along | in from T |
|---|---|
| 0″ | 0 |
| 3″ | 1/8 |
| 6″ | 1/2 |
| 9″ | 1-1/4 |
| 12″ | 2-1/4 |
| 15″ | 3-1/2 |
| 18″ | 5-1/4 |
| 21″ | 7-1/4 |
| 24″ | 9-5/8 |
| 27″ | 12-5/8 |
Off edge S — the steep end
Tick along edge S, the right-hand cut, starting at the factory corner end. Then square in from S.
| along | in from S |
|---|---|
| 0″ | 0 |
| 3″ | 1/8 |
| 6″ | 1/4 |
| 9″ | 5/8 |
| 12″ | 1 |
| 15″ | 1-5/8 |
| 18″ | 2-3/8 |
| 21″ | 3-1/4 |
| 24″ | 4-3/8 |
| 27″ | 5-5/8 |
| 30″ | 7-1/16 |
| 33″ | 8-3/4 |
| 36″ | 10-11/16 |
| 39″ | 12-15/16 |
This table is the longer one on purpose — past the halfway point the curve swings away from T faster than it swings away from S, so the middle of the bean is plotted from here.
Yes — mark on the good face, and sand it off, don’t erase
Good face up, all day, marked in pencil. That is exactly what the reverse-tooth blades bought you: they cut on the down stroke, so the tearing happens on the underside and the face you eat off stays clean. Marking on the back would mean cutting blind to your own line, and it would mirror the whole layout.
Almost none of it survives anyway. Every offset you square off S and T lands in the waste corner — the jigsaw eats them. The only marks left on the finished top are the ticks along the two cut edges and the 24 dots themselves, and all of those sit within a pencil width of the cut line, so the 80-grit fairing pass in Step 8 takes them with it.
Pencil only, and press lightly. A Sharpie bleeds into open grain and ghosts straight through shellac — there is no sanding that out of a face veneer this thin. And don’t bear down with the pencil either: a burnished graphite groove is harder to remove than a line. Skip the eraser too, which just smears graphite into the grain; let the sander do it.
Check the rhythm before you spring the batten
Stand back and read the dot field. Walking the curve from either straight edge toward the middle, the dots should step away from their edge by a little more each time, never by less — and the two runs should meet head-on around the middle of the bean with no kink and no gap wider than the rest. A dot that jumps is a transposed number. Finding it now costs thirty seconds; finding it after the cut costs the sheet.
Same curve, same 23 points, measured from the two edges that become the finished straight sides. Use these as a cross-check on any dot that looks wrong. Use A for the flat run and B for the steep end — near each end the curve turns hard and offsets taken from the wrong baseline get unreliable fast. They meet at 36″ / 27⅞″: if your two dots land on top of each other there, everything is square and the rest will be right.
Table A — from the 42″ edge
Measure across the 42″ edge, then down parallel to the 54″ edge.
| across | down |
|---|---|
| 0″ | 54″ |
| 3″ | 53-7/8″ |
| 6″ | 53-1/2″ |
| 9″ | 52-3/4″ |
| 12″ | 51-3/4″ |
| 15″ | 50-1/2″ |
| 18″ | 48-3/4″ |
| 21″ | 46-3/4″ |
| 24″ | 44-3/8″ |
| 27″ | 41-3/8″ |
| 30″ | 37-3/4″ |
| 33″ | 33-3/8″ |
| 36″ | 27-7/8″ |
Table B — from the 54″ edge
Measure down the 54″ edge, then across parallel to the 42″ edge.
| down | across |
|---|---|
| 0″ | 42″ |
| 3″ | 41-7/8″ |
| 6″ | 41-3/4″ |
| 9″ | 41-3/8″ |
| 12″ | 41″ |
| 15″ | 40-3/8″ |
| 18″ | 39-5/8″ |
| 21″ | 38-3/4″ |
| 24″ | 37-5/8″ |
| 27″ | 36-3/8″ |
Mark each dot with a firm pencil cross, not a dash — you need to see its exact centre from standing height.
Step 4 · 10:50 · 15 min
The lattice has to bow 13.7″ off its own chord over a 68″ span, so it pushes hard inward, toward the corner. That means the brads go on the inside of it — they’re posts resisting the spring, not fasteners.
Set the lattice so one long edge runs through the dots.
Drive a brad snug against the opposite edge at each dot — about 1½″ inside your line, and only ¼″ deep. Just enough to stand up against a sideways push.
Work from one end, bending the lattice onto the next dot as you go. Don’t try to place them all first.
Sight down the batten before you trace. Your eye finds an unfair curve instantly and a tape never will. Nudge any dot that looks wrong — trust the eye over the arithmetic here.
Trace the edge that’s sitting on the dots. Pull the brads.
Nothing gets nailed through the lattice, so there’s no splitting and no pre-drilling. Your 23 pinholes end up 1½″ inside the finished edge — 1/16″ and invisible under shellac.
Step 5 · 11:05 · 5 min
Snap off ~12″ of lattice, drive a brad through it near one end, poke a pencil hole 8½″ along. That’s your trammel.
Don’t measure the centre. Set the pivot down roughly in the corner region and slide it until the pencil just kisses both edges — that point is the centre, by definition. Swing the quarter turn.
Tangency is the only thing that can go wrong here: an arc meeting a straight edge at even a slight angle leaves a kink your eye will catch every time you sit down. Finding the centre by touch guarantees it. And if you’d rather, shove any 16–18″ round thing into the corner until it touches both edges and trace that — the exact radius genuinely doesn’t matter, it’s the one dimension we left adjustable.
Step 6 · 11:10 · 35 min
Saw settings
Never force a curve. If it’s fighting you, you’re pushing.
Cut into the waste
Stay a blade’s width outside the line and sand back to it. Sanding to a line is easy; adding wood back is not.
Start at the flat end near the 54″ edge, where the curve is gentlest, and work toward the steep end.
Relief cuts
Every 6–8″, run a short cut from the outside edge of the sheet in to your line. The offcut falls away in chunks instead of hanging on and binding the blade. Matters most near the 42″ edge where the curve turns hardest.
Corner arc last
Do the big curve first while the panel is at its most rigid, then the little corner. It’s a tight radius for a jigsaw — go slow and let the blade steer.
Step 7 · 11:45 · 15 min
Carry it in rough. If something’s wrong you want to know now, not after fifty minutes of sanding.
Positioning the stone — there is essentially one right spot
The biggest circle this shape can swallow is 38.1″ and your table is 38.0″. It fits with about a twentieth of an inch to spare, so slide it wrong and stone shows.
The rule is easy to hit by eye: the rim just kisses both straight edges — centre 19″ in from each. It has to be 19″, because that is the stone’s radius. Get those two right and the curve looks after itself.
Lay it on the stone and nudge until the rim touches both straight edges. Crouch and look along each edge — you’re looking for the stone to disappear exactly at the plywood line.
Reach underneath with a pencil and trace the stone’s rim onto the plywood’s underside. Everything afterwards gets easy once you can see where the stone actually is.
While it’s up there, mark your three hook positions on that traced line: 19″ along the 54″ edge, 19″ along the 42″ edge, and once on the curve. Those are the three points where rim and plywood edge come flush.
Note where the two tips land. The north tip hangs 11″ past the stone and the west tip 21″ — those are your prop positions.
Step 8 · 12:00 · 50 min
Fair the edge — 80 grit, by hand
The palm sander won’t do this. It’s a face tool. Wrap 80 around a scrap of your 2×4 and work the edge with that — a stiff block, never your fingers, or you’ll deepen the dips instead of cutting the humps.
Sight down the edge with a light behind it. Plywood shows every wobble as a shadow line. Trust your eye over the pencil line.
Keep the 80 off the faces
80 on a palm sander is an edge-and-defect tool. Face veneer on sanded ply is about 1/32″ — a random-orbit sander with 80 goes through that in well under a minute of dwelling, fastest at the edges where the sander rocks, and once you are through there is no fixing it. The faces came sanded from the mill; they want 120 light, then the sponge, nothing more.
Then 120, then the 200 sponge
Edge first with the block, then the faces with the palm sander, going with the face grain. You’re not chasing furniture finish — 200 is plenty under shellac and going finer slightly hurts adhesion. Vacuum between every grit: one 80-grit particle dragged under the 120 pass leaves a scratch you’ll only see once the shellac is on.
Ease the edge — 1/8″ on top, 1/16″ underneath
Run the block at 45° along the edge until the flat facet is about 1/8″ wide on the top face and half that underneath, knock the two corners that facet creates, then round it with the 200 sponge — the sponge flexes around the curve, which is exactly what a flat sander can’t do. Don’t go past 3/16″ on top or the edge starts to read thin.
Two tests beat measuring. Drag the back of a fingernail across it — if it catches or you feel a line, keep going. Press a forearm down for ten seconds — a red line means still too sharp. A raw plywood edge at toddler head height is genuinely sharp, and you have a one-year-old.
There’s a finish reason as well as a safety one: shellac won’t hold on a sharp arris. Film finishes pull away from a knife edge as they cure, so a sharp edge is where the finish fails first and where water finds the core.
Round the two tips — they’re 90° corners
Easy to miss, and worth more than the edge easing. A quarter ellipse meets its straight edges at a right angle, so where the curve lands on the 42¼″ edge, and again where it lands on the 54¼″, you have a sharp point. Toddler-face height, and short-grain plywood points chip when anything knocks them.
Round both to roughly a 1″ radius. Freehand with the sander, no layout — knock it round and fair it by eye. Costs 0.2 square inches of tabletop, takes two minutes.
Preview the pencil with alcohol
120 lifts ordinary pencil easily. The trap is that a line that looks gone on bare wood can come back under shellac — shellac wets the surface and darkens graphite pushed into the grain, and you find out after it is sealed in.
So check first, and the damp wipe-down you were going to do anyway is the test — water darkens residual graphite exactly the way shellac will. Look at the surface while it is wet. The cost is that water raises the grain, so let it dry 15–20 minutes and take one light pass with the 200 sponge before finishing.
Vacuum before you wipe — alcohol on a dusty surface makes mud. Pour onto the cloth, never onto the wood; damp, not dripping. Each pass stays dark 20–60 seconds, and you want to look across the surface with light raking it, not straight down. Do the sawn edge too, which is where pencil and scratches hide. 91% raises no grain and you can finish straight after; 70% is a third water, so give it ten minutes and one light sponge pass.
The better catch here isn’t pencil — it’s swirl marks. The little pigtail scratches a random-orbit sander leaves are invisible on dry bare wood and glaring under a finish, and they are the usual reason a first coat disappoints. Curly scratch patterns mean that spot wants another pass at 120 with lighter pressure.
Rubbing alcohol from the bathroom does it with no grain raise — isopropyl, 70% or 91%, either is fine, and it flashes off in seconds. Worth a ten-second look in the cabinet; it saves you the dry-and-resand. Denatured alcohol (also sold as methylated spirits or shellac thinner) is the traditional answer but there is nothing to buy here. Whatever you use, do it before any shellac — alcohol dissolves the cured film. If a faint line survives, work that spot alone and re-test; don’t take the whole face down chasing it.
A stray brad hole in the good face
Sanding will not remove it — a puncture is not a surface mark, and sanding it out means removing its full depth, far past your 1/32″ of veneer. You would trade a 3/64″ hole for a bald patch.
On a two-day mock-up, leave it. If it nags: a drop of water in the hole, a damp cloth, and the tip of a hot iron for a few seconds swells the fibres most of the way back — a drop, not a puddle, and two applications at most, because soaking plywood lifts face veneer off the glue line. Then dry fully and hit it with the sponge.
Don’t use wood filler under a clear finish; it flashes a different colour and reads as a spot, worse than the hole. If you want it gone properly, a wax fill stick after the shellac cures lets you match the finished tone.
Then clean properly
Vacuum, then wipe with a barely damp cloth, then let it flash off five minutes. Shellac shows every speck of dust you leave, and the cut edge holds more than you think.
Finishing indoors, away from the sanding dust
Don’t shellac it sitting on the stone — it will run off the edge onto the travertine. Put the top up on blocks, paint cans, upturned bowls, anything, on a drop cloth. The edge is the part that most needs sealing and you cannot reach it lying flat.
Open a window, and if you run a fan point it out. A fan blowing across the piece stirs dust onto wet finish. Alcohol vapour is flammable, so no lit burner on the stove while you are laying it on. Let the room settle ten minutes first, and don’t vacuum right before you start — indoor dust is not less than the garage, just different: lint and hair instead of sawdust.
Shellac rags go in the bin normally. The spontaneous-combustion warning belongs to linseed and danish oil, not shellac — but remember it if you ever switch to an oil finish. And you don’t need solvent for cleanup either: bin the cheap brush, or stand it in rubbing alcohol.
Step 9 · 1:00 · 15 min
Rag the faces, brush the edge. A folded cotton rag lays a thinner, more even coat than any brush; the edge is the bit a rag can’t wrap, and it’s also the bit that most needs sealing — bare plywood edge drinks liquid straight into the core.
One pass, then leave it alone. Shellac skins over in under a minute, so going back across a tacky spot tears it. Anything you miss gets fixed by coat two. Hit the cut edge harder than the faces.
Underside: one quick thin coat is worth the two minutes if you have the shellac to spare, but it’s genuinely optional for a two-day mock-up.
Step 10 · 1:15 · 45 min · while it dries
Each hook is two pieces. They live entirely underneath, so nothing lands on the eating surface.
Cut list — times three
| 3 | 2×4 spacer, laid flat | 5″ |
| 3 | ply offcut catch | 6″ |
| 6 | 2″ screws — spacer up into ply | — |
| 6 | 1¼″ screws — catch to spacer | — |
Measure your stone thickness first. A 2×4 flat drops 1½″, which suits stone up to about 1¼″ with shims to spare. Thicker and you want the 2×4 on edge.
How it goes together
Spacer screwed to the plywood underside, just outside your traced rim. Catch screwed to the spacer’s bottom face, reaching ~2″ back in under the stone. Shims driven between catch and stone to snug it.
Countersink the 2″ heads flush — they sit on the spacer’s bottom face and the catch has to lie flat on it. A proud head means a rocking hook. The 1¼″ heads are on the catch’s underside where nothing touches them.
Drill a 1/8″ clearance hole through the 2×4 only. If the screw grips both pieces it jacks them apart instead of pulling them together.
Pad every stone contact — felt, cork, rug-pad scrap. Travertine is soft and spalls at edges. Snug, not cranked.
Step 11 · 2:00 · 20 min
Light pass with 220 — you’re knocking down raised grain, not removing the coat. Wipe the dust. Second coat exactly like the first.
Two coats shrug off a spilled drink for a few minutes, not a whole dinner with four kids. Put a cloth or a vinyl runner down for the meal itself. You’re testing a shape, not a finish.
Step 12 · 2:20 · 30 min · while it dries
Slide the round table into place and lay the non-slip pad over the stone.
Set both props. North tip 11″ past the stone, west tip 21″. A stack of books on a chair seat, a box, a bucket with a board across it — anything rigid that reaches the underside without rocking. Both tips sit at the ends of the bench runs, not the middle, so they’re not where anyone’s legs go.
Tape the bench footprints on the floor while you’re down there: 62″ along the north wall, 61″ down the west, 16″ deep. Without them you’re only testing half the idea.
Step 13 · 3:05 · 30 min
Set the top down, line the traced rim up with the stone, fit the three hooks and drive the shims until each is snug.
The tip test. Press down hard on each tip and watch the far side of the pedestal base. If it lifts off the floor at a load you could apply one-handed, the props aren’t doing their job — adjust until they are. Much easier to feel than to argue about.
Tell the kids once. Don’t lean on, sit on or climb the pointed ends. A propped plywood top is fine to eat off and not fine to treat like furniture. Two days, then it comes off.
Tonight & tomorrow
Carry a full laundry basket from the deck door to the kitchen. Twice. This is the 20.6″ pinch and it’s the thing most likely to make you regret the bean. If it grates, the symmetric 44″ fan gives you 23.9″ back for two seats’ worth of tabletop.
Eat in the real positions — three along the 42″ edge, two down the 54″, one at the corner. Watch whether the corner seat is genuinely usable or only usable on paper.
Leave it two days. Mock-ups flatter themselves on day one; the complaints arrive on day two.
Bolting to the flange — right joint, wrong time
The flange is the correct place to fasten a top: it is the one spot on this table that is already load-bearing, and from the photo it is three accessible hex heads. The mount is not the problem. What leaves with the stone is the problem — that slab is about 123 lb of the 192 lb holding this table down, and it is the only reason a 40″ overhang stands up at all.
Push down on the far tip. This is how much it takes to lift the base off the floor:
| Base foot | Stone on | Stone off | |
|---|---|---|---|
| 14″ | 41 lb | 17 lb | |
| 16″ | 48 lb | 20 lb | |
| 18″ | 56 lb | 23 lb | measure yours |
| 20″ | 64 lb | 27 lb | |
| 22″ | 73 lb | 30 lb | |
| 24″ | 82 lb | 34 lb |
Twenty pounds on the tip is a three-year-old leaning, or a one-year-old pulling himself up on the edge. With the stone in it takes two or three times that. Keep the stone until the base changes.
The stiffness goes too, and that is the part nobody expects. Right now the plywood is bedded on a 38″ disc, so it only cantilevers the last 17″ and barely moves — about 1/16″ under a lean. Bolted to a bare flange it cantilevers 32″, and the same lean gives you 3/8″ of sag. It would feel springy, and 3/4″ ply worked that hard right at the flange edge is where a crack would start.
When you do bolt it — and NOT with the 2″ screws
Those 2″ screws are sized for the hooks: 1½″ of 2×4 plus the top. Through a 1/4″ flange a 2″ screw comes out the eating surface by an inch and a quarter. Length is flange thickness + washer + at most 5/8″ into the ply, which for a 1/4″ flange means a 1″ #12 pan- or hex-head, with a fender washer to span the oversize bolt hole. Your washer instinct was right.
Pilot 9/64″ for a #12, and wrap tape on the bit at 5/8″ — that is the whole safety margin. Mark the holes with the top face-down on the floor and the pedestal set on it, then drill on the bench. Do not drill overhead into a mounted top.
And move the pedestal while you are there. It sits at 19″/19″ only because the stone had to be covered. With the stone gone, put the flange centre 17-7/8″ from the 54¼″ edge and 23″ from the 42¼″ edge — the bean’s balance point. Worst overhang drops 40″ to 36″ and every number in the table above improves by a tenth. It is free.
On “should we have gone bigger”
Bigger makes this worse, not better. What reads as wrong is not the size of the top, it is a round pedestal under a shape that is not round — the mass sits at 19″/19″ while the top reaches 40″ one way and 30″ the other. Grow the top and the overhang grows with it, and the 20.6″ walkway shrinks.
This is exactly what the mock-up was for, and it just told you the answer: the next thing to change is the base, not the bean. A low steel plate cut to follow the bean’s own footprint puts support where the top actually goes, and it is the one piece nobody sells off a shelf. Live on it two days before deciding whether the top also wants to be 44 or 46.